Once you have the format and can convince yourself that you are parsing the data properly, I would use a hash to count tornado days. Something like this:

while (<>) {
   # parse line
   my ($date, $state) = ...
   $tornadoes{$state}{$date}++;
}
foreach my $state (1..50) {
    print "State = $state,   tronadoes = ", scalar keys %{$tornadoes{$
+state}}, "\n";
}
[download]

@STATE_LOOKUP = (
   undef,      # There's no state 0
   'Maine',
   'Hawaii',
   'Canada',
   'New York',
   ...
);

my $data;

open(FILE, ...) or die(...);

while (defined($line = <FILE>)) {
   my ($state_id, $date) = get_fields($line);
   my $date_n = convert_date_to_number($date);
   $data->[$state_id]->{$date}++;
}

close(FILE);

{ # Sample output.
   my ($state_id, $date_n);
   for ($state_id=1; $state_id<=50; $state_id++) {
      my $state = $STATE_LOOKUP[$state_id];
      foreach $date_n (sort { $a <=> $b } keys %{$data->[$state]}) {
         my $date = convert_date_to_string($date_n);
         my $num_tornadoes = $data->[$state]{$date_n};
         printf("State %s had %d tornado(es) on %s\n",
            $state,
            $num_tornadoes,
            $date,
         ) if $num_tornadoes;
      }
   }
}
[download]

get_fields depends on the format of your data.

Look into the Time modules for help implementing convert_date_to_number.

Others will probably use a hash for states without thinking, but an array would be more efficient and just as powerful (assuming you have reports for many states).

As you read though the file, construct a datastructure like this:

%events = ( 1 => { '10/27/2002' => 1,
                   '09/15/2004' => 2,
                   '03/24/2003' => 1 },
            2 => { '02/17/2001' => 1,
                   '12/05/2003' => 2 },
            3 => .....                 );
[download]

That way, as you iterate through the flat file, you'll use state ID to get to the top-level of the hash of hashes, and dates will become the keys for the 2nd level. You don't really care about an event count per date, but as long as you're using hash keys to guarantee unique dates within each state, there's no reason not to use a ++ to increment the event count for that particular date.

Now, you can count event dates in state 4 like this:

my $date_count = keys %{$events{4}};
[download]

It's all about the right datastructure. ...having said that, someone undoubtedly will come up with an even better one. ;)

On the other hand, if you plan to search by date more often than by state, reverse the order so that the top level key is a date, the second level is state.

Update: Just saw the comment that "others will use a hash without thinking." I did think about it, and decided to use a hash because not all 50 states have tornados (ie, a sparse array, which is a good use of a hash). I don't think you'll ever see one in Hawaii, nor in Washington or Vermont. ...at least maybe not within a narrow timeframe of a few years. Also I chose the hash approach on the off chance that as the script gets refined, state numbers might end up becoming state names or state abbreviations, which lend themselves better to hash keys. After discussing it in CB I have to admit that an array will give a more efficient by-state lookup, while the hash will make it easier to sum up the number of states that had an event. Maybe it's a tossup. ;)

If that is the case, think about creating an array of 50 elements (one for each state). The value of each element is a reference to a hash, where the keys of the hash are the dates when one or more tornados occurred in the given state (the value assigned to each element is of no importance: it is the existence of a given date as a hash key for the given state that matters). So, suppose the data is like this:

State  Date        other stuff
1      03/03/2003  one bad mother of a storm...
5      03/05/2003  nothing to write home about...
8      02/08/2003  they said it could never happen here
...
[download]

my @states; # array holding one hash per state

while (<>)
{
   my ($state_number,$date) = split;
   $states[$state_number]{$date} = undef;
}

# now to summarize:
# for each state that had any tornados at all,
#    list number of tornado days:

for ( 1 .. 50 ) {
   if ( defined $states[$_] ) {
      printf( "state_id %2d: %3d tornado days\n", $_,
              scalar keys %{$states[$_]} );
   }
}
[download]

foreach (<MYHANDLE>) {
    $_ =~ /^(\d+)\/(\d+)\/(\d+)\t(\d+)\t/;
    if (!$h{"$1$2$3$4"}) {
        $h{"$1$2$3$4"} = 1;
        $c{$3}{$4}++;
    }
}
foreach $year (sort keys %c) {
    print "$year\n";
    $p = $c{$year};
    foreach (sort {$a <=> $b} keys %$p) {
        print "$_ " . $c{$year}{$_} . "\n";
    }
    print "\n";
}
[download]